The Architecture of Functional Neural Circuits in ... - Semantic Scholar

The Architecture of Functional Neural Circuits in the Vertebrate Retina The Proctor Lecture Helga Kolb U nderstanding the organization of the vertebrate retina has been the goal of many talented visual scientists during the past 30 years. With Cajal's (1892) anatomic descriptions of the cell types that constitute the retina in a number of vertebrate species, and with an early understanding of the role of visual purple in photochemistry in combination with psychophysical studies of adaptation and color vision, we had in the sixties the rudiments of an understanding of how the retina might be organized and functioning. To go further, though, we were beginning to need detailed information of neural circuits that underlay these functions. It was the advent of electron microscopy, microelectrode recording techniques, and pharmacology that then allowed us an era of very rapid advancement. The purpose of this presentation is to summarize these recent advances and to describe our present understanding, based primarily on anatomic investigations, of the underlying architecture of four important functional circuits in the vertebrate retina. The retina is formed during development from an outpouching of the brain that becomes encased in the protective and nourishing eyeball layers behind the focusing structures of the pupil and lens (see refs. 1 and 3 for a review). Light rays forming images of the outside world directly impinge upon this piece of the brain known as the retina, where they stimulate first the sensory rods and cones and then a cascade of neurons to create an output pattern of neural activity in the ganglion cells. All vertebrate retinas are similarly organized into three layers of neurons and two areas of neuropil. Each vertebrate species has some kind of topographic specialization of the retina unique to the species, such as a fovea or a visual streak, predominance of rod or cone photoreceptors, or presence or

From the Departments of Ophthalmology and Physiology and the John A. Moron Eye Center, University of Utah Health Sciences Center, Salt Lake City, Utah. The videotape from which this paper was taken is available from ARVO, 9650 Rockville Pike, Bethesda, MD 20814-3998. Reprint requests: Dr. Helga Kolb, Department of Ophthalmology, John A. Moran Eye Center, University of Utah Health Sciences Center, 50 North Medical Drive, Salt Lake City, UT 84132.

Investigative Ophthalmology & Visual Science, April 1994, Vol. 35, No. 5 Copyright © Association for Research in Vision and Ophthalmology

absence of cone types specific for certain wavelength of stimulation, but the basic cell layers and neuropils are constant. In a vertical section of the human retina as shown in Figure 1, we can see that the retina is a highly complex, layered structure containing millions of closely packed nerve cells. The retina is actually organized in reverse from a way that might be intuitive. The photosensitive elements are not the first structures to be hit by the image. Instead, light rays have to pass through the whole thickness of the retina before stimulating the rods and cones, and, in contrast, the output message to the brain can only exit the retina from the most superficial cell layer containing the ganglion cells. A naive understanding of the retina emphasizes only the sensory photoreceptors and the ganglion cells; however, it is evident on deeper probing that there are many interneurons packed into the central nuclear layer here (Fig. 1) and the neuropil layers, particularly the inner plexiform layer (IPL), are thick and composed of a tangle of interconnecting neural profiles. The first area of neuropil is the outer plexiform layer (OPL), where connections between rod and cones and vertically running bipolar cells and horizontally oriented horizontal cells occur. The second neuropil of the retina, the IPL, functions as a relay station for the vertical information-carrying nerve cells, the bipolar cells, to connect to ganglion cells. Moreover, a multitude of different varieties of horizontally and vertically directed amacrine cells somehow interact in further networks to influence and integrate the ganglion cell signals (Fig. 1). It is only at the culmination of all this neural processing that the message concerning the visual image is transmitted to the brain along the optic nerve. The central question for many of us concerns what kinds of interaction are going on between the cells of the layers of the retina before the visual message is sent to the brain. Clearly, the retina is more than just photoreceptors and ganglion cells, and we need to understand what circuits of neurons are interposed between the photoreceptors and the optic nerve and how they are organized for functional roles.

2385

2386

Investigative Ophthalmology & Visual Science, April 1994, Vol. 35, No. 5

Henle OPL INL IPL

m

tLIGHT ft

GCL NFL 25

ILM

FIGURE 1. Light micrograph of a vertical section through the human retina from an area 2 mm from the foveal center. Ph, photoreceptors; OLM, outer limiting membrane; ONL, outer nuclear layer; Henle, Henle fiber layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer; NFL, nerve fiber layer; ILM, inner limiting membrane; H, horizontal cells; b, bipolar cells; a, amacrine cells. Scale bar, 25

TECHNIQUES THAT HAVE BEEN USED TO UNDERSTAND NEURAL PATHWAYS IN THE RETINA In 1892, the great Spanish anatomist Ramon y Cajal2 first described many of the different nerve cells that make up these neural layers of the retina and contribute processes to the synaptic neuropils. He made beautiful drawings and descriptions of the different nerve

cells in a variety of vertebrate retinas, after staining with a technique invented by Italian anatomist, Camillo Golgi.3 The Golgi silver staining technique had the advantage over other silver techniques in staining individual nerve cells in isolation from surrounding nerve cells. Also, it stains the complete cell throughout its finest dendritic and axonal processes. Cajal was the first to state that bipolar cells of the retina were involved in rod or cone channels of information through

The Proctor Lecture

the vertical pathways and that horizontal and amacrine cells were involved in lateral interactions. He introduced the idea that there were specific connections between specific cell types in the plexiform layers by virtue of stratification patterns of axons and dendrites in the neuropils. Following in the footsteps of Cajal and Steven Polyak,4 who worked on the primate retina, more recent Golgi studies have described the different nerve cell types in retinas that are receiving attention from physiologists and where intracellular recordings can be made.5"10 Cajal's and Polyak's images and classifications of cell types were comprehensive and were often done on retinas that were good experimental models, too; however, they were made from views of vertically sectioned retina. Unfortunately, vertical sections can only give us a limited view of stained neurons because their complete dendritic trees are usually wider than section thicknesses. We need to understand full dendritic tree dimensions and details of dendritic morphology that could indicate synaptic interactions in networks of neurons. This is where wholemount Golgi staining techniques, developed in several laboratories710"14 at about the same time, were a great improvement. Some of the cell types seen by Golgi staining in wholemounted cat retina are illustrated in Figure 2. It is obvious that such views give us complete information on dendritic tree expanse and allow us to classify cells with greater ease. For example, cells Al, A2, All, and A8 have much smaller dendritic trees than cell A17 (Fig. 2). In addition, such wholemount views provide information on characteristics of branching frequency and dendritic field coverage that can be measured either subjectively or objectively with computeraided microscopy. For example, the ganglion cells labeled alpha and beta in Figure 2 have more bushy branching patterns than the cell labeled G19. An overview of the whole cell, as is best seen in wholemounts, has been particularly helpful in revealing the differences between three important ganglion cell types of the human retina that are involved with spatial and color vision, namely, the midget ganglion cells, the small parasol ganglion cells, and the large parasol ganglion cells.46 It is only when the these three cell types are Golgi stained close together in the same area of retina that we see that midget and small parasol cells are two distinct types. Figure 3 shows clearly that the midget ganglion cell has an 18 /urn diameter dendritic tree compared with the small parasol, which is 35 fim in diameter. In contrast, the large parasol cell, lying close by the two previous cells, has an 85 ixm dendritic tree span (Fig. 3). Other authors have, since Polyak's time, confused the midget and small parasol cell types and lumped them together as "midget" ganglion cells.1516 However, the latter classification scheme is too broad a generalization that does not

2387

consider functional morphology, a topic we shall return to later in this paper. A very important aspect of neural architecture in the retina is the arrangement of the neurons' dendrites and axons in a stratified manner. This is best appreciated by through-focusing on stained cells in wholemount preparations or by viewing stained cells in vertical sections. This aspect of neuronal morphology, which had been so emphasized by Cajal,2 was unfortunately neglected in some of the following decades. It was only when the significance of stratification involving midget bipolar cell axons and midget ganglion cell dendrites in monkey retina was appreciated1718 (which we shall discuss later in this paper) that we began to look at all retinas, particularly in the IPL, for stratification patterns for all neurons again. Thus, we saw that ganglion cells in the cat retina had specific stratification levels to put their dendrites into contact with particular stratified bipolar cell axons and amacrine cell dendrites (Fig. 4).19-20 These particular ganglion cells (Fig. 4), now known as beta cell types, come as paramorphic pairs. One variety branches only in distal IPL (area labeled "a"), whereas the other restricts its branches to proximal IPL (area labeled "b"), thereby allowing only synaptic interactions between the beta a type and the flat bipolar and between beta b types and the imagination bipolar (Fig. 4). From an understanding of the different nerve cell types that make up the retina, plus an understanding of where their dendritic trees lie in relationship to each other by virtue of stratification in the IPL, we can begin to make guesses as to which cells could interact. However, seeing stained cell processes lying close together does not necessarily mean that they make synaptic contacts. The only method for determining synaptic interaction unequivocally is to study the neuropils of the retina by electron microscopy (EM). EM has revealed differences between many of the neural elements of the retina on cytological criteria. Thus, we have been able to distinguish rod photoreceptors from cone photoreceptors by differences in outer segment morphology and differences in structure of their synaptic endings in the OPL. Rod spherules can be distinguished by their content of a single synaptic ribbon and a single complex of invaginated profiles where they communicate with bipolar and horizontal cells.21'22 Cone pedicles, on the other hand, clearly have larger synaptic endings that contain many synaptic ribbons and invaginated profiles of secondorder neurons.1823"25 In the IPL where bipolar cells contact ganglion and amacrine cells, we learned, originally from Missotten's24>2b and Dowling's and Boycott's elegant studies,23 that bipolar cells can be recognized by the presence of ribbons at their synapses (Fig. 5, rod BP, cone BP), and amacrine cells can be recognized by the pres-

2388

Investigative Ophthalmology & Visual Science, April 1994, Vol. 35, No. 5

FIGURE 2. Camera lucida drawings of Golgi-impregnated neurons of the cat retina as seen in wholemount views. Al and A2 are narrow-field amacrine cells, and A7/AII and A8 are bistratified narrow-field types. A17, by contrast is a wide-field diffuse amacrine cell, a, j8, and 8 (G19) are ganglion cells types. Small numbers or a/b refer to the strata or sublamina, respectively, of the IPL in which the cell's dendrites ramify. Scale bar, 25 fim.

The Proctor Lecture

2389

FIGURE 3. Light micrographs of each of the three major ganglion cell types of the human retina as impregnated by Golgi procedures, (a) The large parasol cell (lpgc) is in the diffuse cone system and has a large-diameter dendritic tree, (b) Midget ganglion cell (mgc) is a single cone/midget bipolar-connected ganglion cell with a very small dendritic tree, (c) The small parasol cell (spgc) has an intermediate dendritic tree diameter and is in the diffuse cone system. Scale bar, 25 nm.

25 FIGURE 4. Camera lucida drawings of Golgi-impregnated neurons in the cat retina as seen in vertical section to show stratification of axons and dendrites in the IPL. /?a ganglion cell has dendrites branching in distal IPL (sublamina a) to overlap the axon branches of FB bipolar cell. /9b ganglion cells has lower branching dendrites in proximal IPL (sublamina b) to contact axon branches of IB bipolar cell. Bipolar cell RB ends in proximal IPL (sublamina b) overlapping (small arrows) the lower strata of dendrites of a narrow-field bistratified amacrine cell AIL Scale bar, 25 nm.

Investigative Ophthalmology & Visual Science, April 1994, Vol. 35, No. 5

2390 vvw

FIGURE 5. Electron micrograph of neural profiles in the IPL. Rod bipolar (rod BP) and cone bipolar (cone BP) axon terminals make synapses at syhaptic ribbons (r) to two postsynaptic profiles which are either amacrine (A) or ganglion cell (G) dendrites. Frequently, amacrine processes make a feedback or reciprocal synapse to the bipolar cell (arrowhead). All amacrine cells communicate with cone bipolar cell axons by means of gap junctions (open arrows). Scale bar, 0.5 /urn. ence of clusters of synaptic vesicles at a dense membrane specialization in conventional synapses (Fig. 5, Al 7). We also found that certain amacrine cells and, in particular, the one labeled All in Figure 5, communicates by electrical synapses, morphologically described as gap junctions, with neighboring neuronal profiles (in this case All engages in a gap junction, open arrows, with a cone bipolar axon, Fig. 5, cone BP). Interpreting these EM images thus allows us to recognize the various general classes of neurons such as bipolar cell, amacrine cell, and ganglion cell and their synapses. However, they do not tell us exactly what types of bipolar, amacrine, or ganglion cells are involved in these interactions. A better method for resolving this quandary is to examine Golgi-stained neurons by EM. This technique was first used in the goldfish retina by Stell8 to determine photoreceptor contacts of Golgi-stained bipolar and horizontal cells in the QPL. 9111318 - 27 - 32 In the primate retina using the same technique, we learned that rod bipolar cells contacted only rod photoreceptors and that cone bipolar cells contacted only cones (both Cajal and Polyak had thought it possible that some bipolar cell types in mammalian and primate retinas might be mixed, contacting both rods and cones, as in nonmammalian retinas). Furthermore, we saw that rod bipolar dendrites penetrated the rod spherule to end only at synaptic ribbons in central invaginating

positions in mammalian retina, whereas cone bipolar cell dendrites made synaptic contacts with cones in one of two ways. As shown in Figure 6, EM of Golgi-impregnated midget bipolar cell types in monkey retina indicated that one variety, called invaginating midget bipolars, made central invaginating contacts to get their dendrites close to the synaptic ribbons in the cone triads (Figs. 6a, 6b, imb). The other variety, called flat midget bipolars (Figs. 6a, 6c, fmb), had dendritic terminals ending at basal junctions on the surface of the cone pedicle away from the ribbon. Another important finding, confirming Stell's and Missotten's studies,8'24'26 was that horizontal cells made synaptic contacts with rods and cones as lateral elements at the synaptic ribbon triads in monkey retina (Fig. 6a, H). It turns out that this arrangement of horizontal cell dendrites on either side of the ribbon synapse in photoreceptors, plus the bipolar cell dendrites forming the central or basal junction position, is very important. A small local circuit is formed here that influences the flow of information throughout the whole retina (Fig. 7). We know that the photoreceptor neurotransmitter, thought to be glutamate, is released in the dark in the vertebrate retina'•33-34 (for reviews). When a light stimulates the photoreceptor, transmitter release ceases and the photoreceptor responds with a

2391

FIGURE 6.

(a) Electron micrograph of synaptic contacts {known as triads) in a cone pedicle of monkey retina. Second-order neurons contact cones as invaginating central elements (imb) at basal junctions (fmb, arrows) or as lateral elements (H) to the synaptic ribbon, (b) A single dendritic terminal of the imb (Golgi-stained profile) projects into the central position of the ribbon triad, (c) Two Golgistained fmb terminals end on the cone pedicle surface at basal junctions (arrows) on either side of the central invaginating dendrite at the ribbon triad. Scale bar, 0.5 /un.

slow, sustained hyperpolarizing response (Fig. 7b), but the postsynaptic bipolar cells respond with either slow, sustained hyperpolarization (Fig. 7c) or slow, sustained depolarization (Fig. 7d) of the membrane.3D~38 The horizontal cells responds, too, with slow,sustained hyperpolarizations (S-potentials),39'40 but a lateral inhibitory feedback synapse from the horizontal cell to the photoreceptor is also thought to occur.83 The feedback signal consists of summed information from a network of horizontal cells connected over a wide spatial area of the OPL. Thus, this large spatial area influences both the photoreceptor and the bipolar cell to include an inhibitory response coming from a surround region of retina.35'38'41'42 This local circuit provides the bipolar cell with a spatially antagonistic receptive field in a center-surround organization. A similar local circuit at the bipolar to amacrine or ganglion cell synapse occurs in the IPL. In Figure 5, we see the manner in which rod and cone bipolar cells are presynaptic at ribbon synapses to two postsynaptic elements. One of the postsynaptic amacrine cell pro-

files in each case makes a feedback (reciprocal) synapse to the bipolar cell (Fig. 5, arrowheads) in the same strategic position that the horizontal cell dendrite does at the photoreceptor ribbon synapse. Some researchers interpret this to indicate a possible similarity in function. Thus, it has been suggested that another surround mechanism coming from the amacrine cells could be added at this synapse in the IPL.4144 A real understanding of even small local circuits, such as those illustrated in Figures 5 and 7, are the result of large-scale serial section analyses. By making tracings of serial electron micrographs, we can reconstruct portions of retinal cells or even the complete dendritic tree of a cell when it is small enough, to study all the synaptic input and output relationships. One of our earliest reconstructions in the cat retina was of an All amacrine cell.45 This small-field amacrine cell type could be reconstructed from its cell body through its primary dendrites and into the finest dendritic terminals. All amacrine cells have a bistratified morphology (Figs. 4 and 8, All), with their distal fine dendrites

2392

Investigative Ophthalmology & Visual Science, April 1994, Vol. 35, No. 5

photoreceptor hyperpolarizing

/c

invaginating bipolar depolarizing

basal contact bipolar hyperpolarizing FIGURE 7. (a) Electron micrograph of the ribbon triad in a cone pedicle in the monkey retina, (b) The cone itself gives a hyperpolarizing slow potential response to a light flash, (c) The bipolar that invaginates to the cone ribbon gives a depolarizing slow potential response to a light flash, (d) The bipolar that contacts the cone at a basal junction gives a hyperpolarizing slow potential response to a light flash. H, horizontal cell dendrites; imb, invaginating midget bipolar cell; fmb, flat midget bipolar cell; curved arrows, possible sites of horizontal cell feed back synapses. Scale bar, 0.5 //m.

postsynaptic to rod bipolar cell axons (rb) and their proximal lobular appendages making chemical synapses to ganglion cell dendrites (Fig. 8, synapses to GCa). Gap junctions join the larger All dendrites with cone bipolar axons (Fig. 8, large black spots, gj to cb). By means of the latter contact, rod-driven information passes through the All channel to the cone bipolar (cb), and thence to the ganglion cell that the cone bipolar synapses upon (Fig. 8, GCb). 45 ' 46 All amacrine cells are now recognized as being pivotal components of the rod system pathways in all mammalian retinas thus far studied, including human retina.G47~51 Another technique that we made use of during the last decade is anatomic investigation of intracellularly recorded and horseradish peroxidase-stained (HRP)

neurons. This technique, which was first used in the retina by Lasansky,52 allows us to determine physiological responses of a neuron to light stimulation, and then to study the synaptic relationships of the same electron-dense marked neuron by EM. It has been used with great success in turtle, fish, rabbit, cat, and even in monkey retina 53 (for review). Figure 9 shows the morphology and synaptology of a small ganglion cell in the cat retina, known as a beta cell, that has been stained with intracellular iontophoresis of HRP. Its appearance by light microscopy in a vertical thick section is shown (Fig. 9a). The cell's light response consists of a burst of spikes at light OFF: it is thus an OFF-center physiological type 43 ' 54 (Fig. 9a). When the stained beta cell is sectioned for EM, the

2393

The Proctor Lecture

GCa

FIGURE 8. Drawing of a three-dimensional ivconsti in lion from serial electron micrographs of an AH ama< i ine cell and its synapiic relationships with rod (rli) and com1 bipolar i ells (clO and ganglion < ells ((Ida, (ICh) in I he I PL of I he maininalian retina, gj, gap junction. See text for further details.

electron-dense HR1' deposit throughout the tell allows it to be readily picked out and its synaptic input studied. In this cast*, synaptic input tomes from cone bipolar cells (Fig. 9b), the All amacrine cell (Fig. 9c), and some other amaciine cell types that we call A8 and A2 or A.S (Figs. 9d, 9e). It was possible to reconstruct the complete dendritic tree of this relatively small beta ganglion cell (Fig. 9a), enabling us to evaluate the total number and distribution of bipolar and amaciine cells impinging on it.4'1 We calculated that this small, central beta cell in cat retina received about 1000 synapses contributed by 12 to 14 bipolar cells, 7 to 10 All amaciine cells, and 28 to 4 1 other ainact ine cell types. Based on what we have learned from using the techniques just described, we are now in a position to understand the architecture of several functional circuits in the vertebrate retina. The remainder of this paper will summarize what we currently consider to be the organization of (1) circuitry of rod pathways underlying night vision, (2) circuits mediating successive contrast otherwise known as ON and OFF pathways, (M) circuits mediating simultaneous contrast or center-

surround organization and, finally, (4) circuitry of red and green color-opponent pathways in human retina.

CIRCUITS OF ROD PATHWAYS UNDERLYING NIGHT VISION EM studies of bipolar cell connections with photorcceptors in monkey and cat retinas 1 8 "'' 1 ' revealed that rod pathways through the retina were differently wired than were cone pathways. Rod pathways involve huge convergences and divergences of cell contacts. They start with many rods converging information to the rod bipolar cell in the Ol'I.. Rod responses are carried to the next stage of processing in the ll'L, where divergence occurs, to several different ainacrine cell types, the most important of which are the All and the Al 7 cells. 7 1 "" '" Rod bipolar cells do not contact ganglion cells directly. Instead, they contact amacrine cells, which spread out (he rod information before converging on ganglion cells. Two other cell types are also influential upon the rod pathways. These are the dopaininergic amacrine" ''" and the (iABAer-

2394

Investigative Ophthalmology 8c Visual Science, April 1994, Vol. 35, No. 5

0-5

A8

All

e FIGURE 9. Electron microscopy of the synapses upon an HRP and intracellularly recorded j3a ganglion cell in the cat retina, (a) Camera lucida drawing of the HRP ganglion cell in vertical section. The cell's dendrites branch in sublamina a. The cell's light response is a depolarization and burst of spikes at the "off" of the light (black bar). Scale bar, 30 fim. (b) Cone bipolar input (CB) recognizable by the synaptic ribbon, occurs to the HRP-stained beta cell dendrite (c). All amacrine cells make a typical punctate synapse (arrows) upon the HRP-stained dendrite. (d) A dark A8 amacrine cell profile and (e) paler A2—A3 profiles also make synapses upon the HRP-stained ganglion cell (arrow). Scale bar, 0.5 pm.

The Proctor Lecture

gic interplexiform cell.6162 They may modulate and feed back and forth rod signals to further amplify and reset levels of dark adaptation between the input in the OPL and output of the system in the IPL. The huge convergence of the rod system has recently been calculated by Sterling and coauthors63'64 and Kolb and Nelson.43 Thus, about 1500 rods have input to a single small-field ON-beta ganglion cell via 100 rod bipolars, five All amacrine cells, and four cone bipolar axons. In the case of a large-field OFF-alpha ganglion cell, 75,000 rods drive 5000 rod bipolars and 250 All amacrine cells before converging on the ganglion cell. In terms of divergence of the rod system, we know that a single rod photoreceptor passes information to two rod bipolar cells, and then through five All amacrines to eight cone bipolar cell axons to drive two ON-beta cells.63'64 So it is that divergent and then convergent circuitry provides pooling and amplification of the rod signal in very low light levels to allow our visual system to be sensitive to a single quantum of light.65 Combining the physiological responses we now understand for the neurons of the rod system in the cat retina,48'66"1'8 with their anatomic connections,46'15 l b l ( ' 9 we can construct a summary diagram (Fig. 10). Signals from hyperpolarizing rod photoreceptors depolarize rod bipolar cells (Fig. 10, rb, physiological response to the left). Rod bipolar cells synapse upon two depolarizing amacrine cells, the wide-field A17 (Fig. 10, A17, response to the left) and the small, bistratified All amacrine cells (Fig. 10, All, response to the right). Rod signals reach OFF-center ganglion cells (Fig. 10, a-type beta GC) via chemical synaptic input from the All amacrine cell lobular appendages (Fig. 10, boxed region a). ON-center ganglion cells (Fig. 10, b-type beta GC) receive rod signals, via All amacrine cell electrical synapses at gap junctions to cone bipolar cells (Fig. 10, ibc, boxed region b). The latter cone bipolars stimulate the ON-center ganglion cells by excitatory chemical synapses. Dopaminergic amacrine cells (Fig. 10, DA) influence the rod pathways by chemical synapses in the IPL upon All and A17 amacrine cells and via an internuncial GABA-ergic interplexiform cell,62'70 which in turn synapses upon rod bipolar cells in the OPL (Fig. 10, IPC).62 CIRCUITS MEDIATING SUCCESSIVE CONTRAST (ON AND OFF PATHWAYS) Cone pathways are less convergent than rod pathways in the mammalian retina. Typically, a relatively small number of cones drive cone bipolars, and a small number of cone bipolars drive a single ganglion cell. In fact, animals with very high visual acuity, like monkeys and humans, have developed the least convergent cone pathways of all, known as midget pathways.4

2395

Cone pathways in mammals and humans run as two parallel streams of information directly from the cone photoreceptor to the ganglion cell through the straight pipeline, the cone bipolar cell.20'56'69 71'72 What is the reason for two parallel channels for the cone system when the rod system had only one? The answer is that this organization allows one channel to provide information to the ganglion cell concerning brighterthan-background stimuli (the ON-center channel) and the other, darker-than-background stimuli (the OFFcenter channel).38'73'74 This is the basis of successive contrast in visual perception.74 The anatomic substrate for the origins of these two important ON-center and OFF-center channels in the bipolar cell was discovered to be the types of synaptic contacts midget cone bipolars made with cone pedicles in the monkey retina.1718 Thus, an EM study of the Golgi-impregnated midget bipolar cells had described one type of midget bipolar cell making invaginating contacts and another making basal contacts with the cone pedicle (summarized in Fig. 6). Rod bipolar cell dendrites were known to make only the invaginating type of contact with rod spherules.18 It is now thought that in the monkey retina, postsynaptic midget bipolar cells respond to light just like the photoreceptor with a slow hyperpolarization when the bipolar dendrite ends at a basal junction.75 In contrast, the other type of midget bipolar cell, that invaginates to the ribbon synapse, responds to light with an inverted sign compared with the cone. It gives a slow depolarizing response.75 Thus, the nature of the postsynaptic membrane channels on the cone bipolar cell dendrite is the important designator of the response sign the bipolar will have.36'37'76 The hyperpolarizing bipolar types are the start of OFF-center channels, and the depolarizing types are the start of ON-center channels through the retina. The same difference of contacts with cones has been found for flat and invaginating diffuse cone bipolar cell in the cat retina, too.27 In other words, it is not just a trait of the midget bipolar system in primates. At the level of information transfer between cone bipolar cells and ganglion cells in the IPL, only excitatory channels are present.77 Therefore, the type of signal transmitted by the ganglion cell, as either ON- or OFF-center, is essentially determined by the nature of the cone bipolar cells contacting it. To keep the ON and OFF channels separate through the ganglion cells to the brain, the IPL is essentially divided into two functionally discrete sublaminae, called a (the two strata of the IPL below the amacrine cell bodies) and b (the other three strata of the IPL stretching to the ganglion cell bodies)45 (Fig. 4). Interactions are only allowed between basal-contacting cone bipolars and one set of ganglion cells in sublamina a (Fig. 4, FB and /3a ganglion cells), whereas

2396

Investigative Ophthalmology 8c Visual Science, April 1994, Vol. 35, No. 5

type beta X5C FIGURE 10. Summary diagram of the neurons and their responses involved in rod-driven pathways of night vision in the mammalian retina, rb, rod bipolar; A17, A17 wide-field diffuse reciprocal amacrine cell; ibc, invaginating bipolar cell; All, All bistratified rod amacrine cell; a-type beta GC, small-field OFF-center X ganglion cell; b-type beta GC, small field ON-center X ganglion cell; IPC, interplexiform cell; DA, dopaminergic amacrine cell; a, square, synaptic contacts between All lobular appendages and a-type (OFF-center) ganglion cells; b, square, electrical synapses (gapjunctions) between All dendrites and ibc axon to pass signals to b-type (ON-center) ganglion cells. See text for full description.

invaginating-contacting cone bipolar cells can only interact with another set of ganglion cells branching in sublamina b (see Fig. 4, IB and fib ganglion cells).19 Gouras was the first to suggest that this specificity of bipolar-to-ganglion cell contacts underlay ON-center and OFF-center midget ganglion cell responses in monkey.17 Later, Nelson conclusively proved this hypothesis by means of intracellular recording and marking experiments in ganglion cells of cat.78 Figure 11 shows the intracellular recordings from two different ganglion cells of Nelson and coauthors' study.78 One cell proves to be ON-center, giving a burst of spikes riding on a depolarization of the membrane as soon as the light flash goes on (Fig. 11). In contrast, the other cell is OFF-center, giving a hyperpolarization to the membrane when the light flash is on but a burst of spikes riding on the depolarization when the light flash is over (Fig. 11). The ON-center ganglion cell has dendrites restricted to branching in sublamina b. The OFF-center cell has dendrites reaching higher to branch only in sublamina a (Fig. 11). A summary of the manner in which the cone pathways carry the message concerning brightness and

darkness through the retina in the cat is shown in Figure 12. Cones hyperpolarize to light, but two bipolar channels, one carried by a depolarizing bipolar (Fig. 12, ibc and light response) and the other by a hyperpolarizing bipolar (Fig. 12, fbc and light response), split the original cone signal into lightness or ON-center and darkness or OFF-center. These bipolar responses are transmitted directly to ganglion cells architecturally separated to the different sublaminae of the IPL, resulting in one channel of ganglion cells with dendrites in proximal IPL (sublamina b) becoming ON-center and the other type with dendrites only in distal IPL (sublamina a) becoming OFF-center (Fig. 12). CIRCUITS MEDIATING SIMULTANEOUS CONTRAST (CENTER-SURROUND RECEPTIVE FIELDS) Information concerning the overall brightness or darkness of the image is of primary importance for visual sensation, but putting them in simultaneous contrast to each other greatly improves the resolution of the

2397

The Proctor Lecture

35 10mV

ON-center

OFF-center

FIGURE ll. Camera lucida drawings of two ganglion cells that have been intracellularly recorded and stained. The ganglion cell (GC) to the left gives an ON-center burst of spikes to a light flash, and its dendrites branch in sublamina b of the IPL. The ganglion cell (GC) to the right gives an OFF-center response to light, and its dendrites branch in sublamina a. Scale bar, 35 pm and 10 mV.

FIGURE 12. Summary diagram of the neurons and their physiological responses to light for the diffuse cone pathways responsible for successive contrast (ON and OFF) pathways in the mammalian retina, ibc, invaginating cone bipolar cell; fbc, flat cone bipolar cell; a-type and b-type CC; OFF-center and ON-center ganglion cells, respectively. See text for full description.

2398

Investigative Ophthalmology & Visual Science, April 1994, Vol. 35, No. 5

image. Simultaneous contrast is achieved by lateral inhibition where a dark boundary inhibits a light area or vice versa. In the retina, lateral inhibition creates a concentric (center-surround) arrangement of the cell's receptive field.79'80 High acuity and color channels particularly need to have a center-surround organization.81 It is thought that horizontal cells at the OPL provide, through a mechanism of lateral inhibition, a surround arranged around the receptive field center of the bipolar cell response.35'38'82 The wiring responsible appears to start at the small, local circuit we saw in the cone triads at the ribbon synapses (Fig. 7). Thus, the negative feedback synapse of the horizontal cell to the cone photoreceptor at the ribbon triad synapse (Fig. 7, curved arrows) allows the larger receptive field of the coupled horizontal cell network to provide a surround to the narrow-field central cone response.4283 This concentric organization is then transmitted to the bipolar cells making contact with the cone. Figure 13 summarizes the architecture for center surround organization. The center pathway is created by the cone to bipolar-to-ganglion cell through-channel, whereas the injection of horizontal cell information provides an antagonistic surround to

the center: an OFF-surround for the ON-center channel (Fig. 13, horizontal cell, and ibc, lefthand pathway) and an ON-surround for the OFF-center channel (Fig. 13, horizontal cell, and fbc, righthand pathway). In cat retina, surround responses are not as strong a component of bipolar cell receptive field as they are in coldblooded vertebrate bipolars.75 However, surround responses are strong in cat ganglion cells.2580-84 Thus, it has recently been suggested that an additional surround is constructed by certain amacrine cells in the IPL in this and other species.43'44

CIRCUITS UNDERLYING RED AND GREEN COLOR OPPONENCY IN THE HUMAN RETINA Color vision is based on the occurrence of at least three different types of cone photoreceptor in the retina, each with a different sensitivity of its visual pigment to different parts of the spectrum.85"88 Thus, there are blue-sensitive, green-sensitive, and red-sensitive cones. It appears certain, then, that three types of cone are the start of color-specific channels in the primate retina.8990

Horizontal

O ON-center OFF-surround

OFF-center ON-surround

FIGURE 13. Summary diagram of the neurons involved in simultaneous contrast or centersurround organization of the receptive field of ganglion cells in the mammalian retina. The center response is formed by the cones to bipolar-to-ganglion cell channels (central spots in receptive field maps) and the concentrically organized antagonistic surround (rings around the central spots of the receptive field maps) is formed by horizontal cells, ibc, invaginating cone bipolar cell; fbc, flat cone bipolar cell; a-type and b-type GC; OFF-center and ONcenter ganglion cells, respectively. See text for a full description.

The Proctor Lecture

From the time of Polyak,4 we have known that primate retinas have a unique set of neurons devoted to single cone, high-acuity, and color-opponent channels not found in species with poor color vision, such as the cat. The diffuse cone channel bipolar and ganglion cells are, of course, also present in primate retinas.61891"94 The latter ganglion cells are the small and large parasol ganglion cells mentioned earlier in this paper (Fig. 3). The exclusive primate midget bipolar and midget ganglion cells have such tiny dendritic trees that they fit single cone pedicles, in the case of midget bipolar cells,18 and single midget bipolar axons in the case of midget ganglion cells.95 Because these midget channels are related to a single cone in the foveal region of the primate retina, and the individual cones are of specific wavelength-preferring types, we know that a single color preference is carried by each midget system neuron throughout the retina. The organization of color channels in foveal ganglion cells of the monkey retina was described by Gouras in 1968.9b The receptive fields of midget ganglion cells fell into two basic types. They were ON- or OFF-center responding and could be stimulated most strongly by either red or green light. Thus, midget ganglion cells were red ON or OFF or green ON or OFF types. Ganglion cells sensitive to blue were also seen, but they appeared to be different and had larger receptive field characteristics. The red and green midget ganglion cells demonstrated sensitivity in their surround to the complementary color of their centers. This made them color opponent. Thus, red center ganglion cells could be red ON-center/green OFFsurround or red OFF-center/green ON-surround. The same organization prevailed for green center ganglion cells. Blue center ganglion cells appeared to have both larger receptive fields and coextensive yellow surrounds.89 Blue-carrying channels do not seem, therefore, to be midget channels. We think that red and green channels are organized through the retina in the manner shown in Figure 14. A red cone would contact two red midget bipolars and through them, two red midget ganglion cells. For example, a red ON-center midget bipolar would contact a single red cone at invaginating synapses in the OPL, and a single red ON-center midget ganglion cell in sublamina b of the IPL. The same red cone would also be contacted by a single red OFFcenter midget bipolar at basal junctions, and the bipolar would, in turn, contact a single red OFF-center midget ganglion cell in sublamina a of the IPL (Fig. 14). Thus, red ON- and red OFF-center receptive field ganglion cell types are generated (Fig. 14, red + center and orange — center, left). A green cone would be connected in a similar manner to two green midget bipolars (ON and OFF types) and two midget ganglion cells, which would also be ON- and OFF-center recep-

2399

tive field types (Fig. 14, dark green + center and light green — center, right). Horizontal cells and small-field amacrine cells are thought to provide the spatially and spectrally antagonist surround response to the four types of receptive field centers (Fig. 14, green and red annuli around the centers).97"99 Two of the three types of horizontal cell could make green surrounds to red center bipolars and red surrounds to green center bipolar cells (Fig. 14, HII and HIII). In addition, red surrounds for a green center ganglion cell might be provided by an amacrine that collects information from a cluster of red-driven midget bipolar cells to pass to a green-driven midget bipolar/midget ganglion cell synaptic complex (Fig. 14, red and orange a). Of course, the opposite of a green-driven amacrine cell providing green surrounds to red center bipolar and ganglion cells could also occur. ARCHITECTURE OF FUNCTIONAL PATHWAYS IN THE HUMAN RETINA During the past 25 years, we have had an explosion in our knowledge concerning the architecture of many of the functional pathways underlying processing of the visual image in the retina. Most of the new knowledge comes from anatomic and electrophysiological investigation of animal models, particularly from investigations of the cat and monkey retinas. Anatomic investigations of human retina have progressed more slowly for obvious reasons, but these investigations have consistently shown that the neurons composing the human retina are so similar in morphology and architecture that we are safe in assuming that they fit into the same circuits we have elucidated in the cat or monkey retina. The culmination of the present knowledge concerning cat and monkey retinal organization as applied to the human retina is shown in the final color schematic of this paper (Fig. 15). This figure depicts a vertical section of human retina containing the cell types and major pathways that have been discussed in this paper. At night, dim light information flows from rods (r) to rod bipolars (rb) to several amacrines, chief among them the small-field bistratified Alls and the wide-field A17s. OFF-center ganglion cells are driven by direct synaptic input from All cells and ON-center ganglion cells by indirect gapjunction mediated All to cone bipolar inputs (Fig. 15, All to di to ON-center g). In daylight, diffuse cone pathways consist of cones contacting two straightthrough channels of bipolar cells (di and df) to ONcenter and OFF-center ganglion cells. ON- and OFFcenter responses of the ganglion cells are determined by specific cone bipolar contacts with cone pedicles, and separation of synaptic interactions between bipolars and ganglion cells to the correct sublamina of the

2400

Investigative Ophthalmology & Visual Science, April 1994, Vol. 35, No. 5

mk. A17

S fa

IPC

A8

2401

The Proctor Lecture FIGURE 14. Summary diagram of the neurons and their receptive fields that are involved in red and green color opponency in the human retina. Lighter colored neurons are ON-center carrying channels and darker colored neurons are OFF-center carrying channels. See text for a full description. HI, HII, and HIII, three horizontal cell types; mb, midget bipolar cells; a, small-field amacrine cells; mgc, midget ganglion cells. FIGURE 15. Summary diagram of a schematic human retina with the neurons of the four major pathways described in this paper shown: (1) night vision [rod] pathways, (2) successive contrast [ON and OFF] pathways, (3) simultaneous contrast [center-surround] pathways, and (4) red and green color-opponent and spatially opponent pathways. Blue pathway is also shown, c, cones; r, rod; HC, horizontal cell; rb, rod bipolar cell; bb, blue-cone bipolar cell; fm, flat midget bipolar cell; im, invaginating midget bipolar cell; di, diffuse invaginating cone bipolar cell; df, diffuse flat cone bipolar cell; A17, A17 amacrine cell (rod system); All, All bistratified rod amacrine cell; A8, A8 amacrine cell, rod and cone system amacrine cell; sfa, small field amacrine cell; wfa, wide-field amacrine cell; IPC, interplexiform cell; mg, midget ganglion cell; SPg, small parasol ganglion cell; LPg, large parasol ganglion cell. ON and OFF, example ON-center and OFF-center ganglion cells of the small parasol types. See text for a full description.

IPL. Spatially opponent surrounds to the ganglion cell responses are provided by horizontal cells (HC) in the OPL and possibly by amacrine cells (for example, A8) in the IPL. Color-opponent pathways for red and green are via the two midget bipolar cells (im and fm) and their specific connections with ON- and OFFcenter midget ganglion cells (mg). Chromatically opponent surrounds could be provided at the OPL by horizontal cells (HC) and at the IPL by small field amacrine cell circuits (sfa). Wide-field amacrine cells (wfa) and interplexiform cells (IPC) are some of the many other interneurons in the retina probably responsible for other integrations, such as coding movement in the periphery and for resetting the retinal pathways in different states of adaptation. More pathways than can be described in the present paper are being discovered at a rapid rate. But so far we only understand some of the simplest pathways and in very simple terms. Undoubtedly, many more details will be resolved in the future, and the day is not too far off when we will understand how this myriad of neural circuitry in the retina codes all the information presented it in the visual scene. Acknowledgments This work has been supported over the years by NEI grants EY03323 and EY04845 and by grants from Fight for Sight and Research to Prevent Blindness. Peter Ahnelt, University of Vienna, Austria, Patricia Goede, Ophthalmology Department, Utah, and Robert McDermott, Supercomputing Institute, Utah, gave unstinting time and effort to the creation of the videotape presentation. I thank Mike Milocek, Supercomputing Institute, Utah, for help with the photography. References 1. Dowling JE. The Retina: An Approachable Part of the Brain. Cambridge: Harvard University Press; 1987. 2. Cajal SR. The Structure of the Retina. (1892) Thorpe

10.

11. 12.

13. 14.

15.

SA, Glickstein M, trans. Springfield, IL: Thomas; 1972. Golgi C. Sullafinaanatomie degli organi centrali del sistemi nerviosa. Rev. Sper. d. Freniat. 1882;665:36. Polyak SL. The Retina. Chicago: University of Chicago Press; 1941. Kolb H. The morphology of the bipolar cells, amacrine cells and ganglion cells in the retina of the turtle Pseudemys scripta elegans. Philos Trans R Soc Lond B. 1982;298:355-393. Kolb H, Linberg KA, Fisher SK. The neurons of the human retina: A Golgi study. / Comp Neurol. 1992;318:147-187. Kolb H, Nelson R, Mariani A. Amacrine cells, bipolar cells and ganglion cells of the cat retina: A Golgi study. Vision Res. 1981; 21:1081-1114. Stell WK. The structure and relationships of horizontal cells and photoreceptor-bipolar synaptic complexes in goldfish retina. Am J Anat. 1967; 121:401424. Stell WK, Lightfoot DO. Color-specific interconnections of cones and horizontal cells in the retina of the goldfish./ Comp Neurol. 1975b; 159:473-501. Stell WK, Witkovsky P. Retinal structure in the smooth dogfish, Mustelus canis: Light microscopy of photoreceptor and horizontal cells, f Comp Neurol. 1973; 148:33-46. Boycott BB, Kolb H. The horizontal cells of the rhesus monkey retina. J Comp Neurol. 1973; 148:115-140. Famiglietti EV. 'Starburst' amacrine cells and cholinergic neurons: Mirror-symmetric ON and OFF amacrine cells of rabbit retina. Brain Res. 1983; 261:138144. Mariani AP. A diffuse, invaginating cone bipolar cell in primate retina. J Comp Neurol. 1981; 197:661-671. Ogden TE. The morphology of retinal neurons of the owl monkey, Aotes. J Comp Neurol. 1974; 153:399428. Dacey DM, Petersen MR. Dendritic field size and morphology of midget and parasol ganglion cells in the human retina. Proc Natl Acad Sci USA. 1992; 89:96669670.

2402

Investigative Ophthalmology & Visual Science, April 1994, Vol. 35, No. 5

16. Rodieck RW, Binmoeller KF, Dineen JT. Parasol and midget ganglion cells of the human retina. J Comp Neurol. 1985; 233:115-132. 17. Gouras P. The function of the midget system in primate color vision. VisionRes. 197l;3(suppl):397-410. 18. Kolb H. Organization of the outer plexiform layer of the primate retina: Electron microscopy of Golgi-impregnated cells. Philos Trans R Soc Lond Biol. 1970;258:261-283. 19. Famiglietti EV, Kolb H. Structural basis for ON- and OFF-center responses in retinal ganglion cells. Science. 1976; 194:193-195. 20. Kolb H, Famiglietti EV. Rod and cone pathways in the retina of the cat. Invest Ophthalmol. 1976; 15:935-946. 21. Sjostrand FS. The ultrastructure of the retinal rod synapses of the guinea pig eye. / Appl Phys. 1953;24:1422. 22. Sjostrand FS. Ultrastructure of retinal rod synapses of the guinea pig eye as revealed by three-dimensional reconstruction from serial sections. J Ultrastruct Res. 1958;2:122-17O. 23. DowlingJE, Boycott BB. Organization of the primate retina: Electron microscopy. Proc R Soc Lond Biol. 1966;166:80-lll. 24. Missotten L. The Ultrastructure of the Human Retina. Brussels: Arscia Uitgaven; 1965. 25. Rodieck RW. The Vertebrate Retina: Principles of Structure and Function. San Francisco: WH Freeman; 1973. 26. Missotten L. Etude des synapses de la retine humaine au microscope electronique. Proc Eur Reg Conf Electron Microsc (Delft). 1960; 2:818-821. 27. Boycott BB, Kolb H. The connection between the bipolar cells and photoreceptors in the retina of the domestic cat. J Comp Neurol. 1973; 148:91-114. 28. Kolb H. The connections between horizontal cells and photoreceptors in the retina of the cat: Electron microscopy of Golgi-preparations. f Comp Neurol. 1974;155:1-14. 29. Kolb H, Jones J. Electron microscopy of Golgi-impregnated photoreceptors reveals connections between red and green cones in the turtle retina, fNeurophysiol. 1985; 54:304-317. 30. Mariani AP. Giant bistratified bipolar cells in the monkey retina. Anal Rec. 1983;206:215-220. 31. Mariani AP. Bipolar cells in monkey retina selective for cones likely to be blue-sensitive. Nature. 1984b; 308:184-186. 32. Stell WK, Lightfoot DO, Wheeler TG, Leeper HF. Goldfish retina: Functional polarization of cone horizontal cell dendrites and synapses. Science. 1975; 190:989-990. 33. Bortoff A. Localization of slow potential responses in the Necturus retina. VisionRes. 1964;4:627-635. 34. Massey SC. Cell types using glutamate as a neurotransmitter in the vertebrate retina. Prog Ret Res. 1990;9:399-425. 35. Kaneko A. Physiological and morphological identification of horizontal, bipolar and amacrine cells in goldfish retina./Physiol. (Lond). 1970;207:623-633. 36. Slaughter MM, Miller RF. 2-amino-4-phosphonobu-

37.

38.

39.

40.

41. 42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

tyric acid:A new pharmacological tool for retina research. Science. 1981; 211:182-184. Slaughter MM, Miller RF. An excitatory amino acid antagonist blocks cone input to sign-conserving second-order retinal neurons. Science. 1983; 219:12301232. Werblin FS, DowlingJE. Organization of the retina of the mudpuppy, Necturus maculosus: II: Intracellular recording. / Neurophysiol. 1969; 32:339-355. Svaetichin G, MacNichol EFJ. Retinal mechanisms for chromatic and achromatic vision. Ann NY Acad Sci. 1958; 74:385-404. Tomita T. Electrophysiological study of the mechanism subserving color coding in the fish retina. Cold Spring Harbor Symp. Quant Biol. 1965;30:559-566. Naka K-I. Neuronal circuitry in the catfish retina. Invest Ophthalmol. 1976; 15:926-935. Toyoda J-I. Membrane resistance changes underlying the bipolar cell response in the carp retina. Vision Res. 1972; 12:283-294. Kolb H, Nelson R. Off-alpha and off-beta ganglion cells in the cat retina: II: Neural circuitry as revealed by electron microscopy of HRP stains./ Comp Neurol. 1993;329:85-110. Marc RE, Liu W-S, Muller JF. Multiple GABA-mediated surround channels in goldfish retina. Invest Ophthalmol Vis Sci. ARVO Abstracts. 1988; 29:272. Famiglietti EV, Kolb H. A bistratified amacrine cell and synaptic circuitry in the inner plexiform layer of the retina. Brain Res. 1975;84:293-300. Kolb H, Famiglietti EV. Rod and cone pathways in the inner plexiform layer of the cat retina. Science. 1974; 186:47-49. Chun M-H, Han S-H, ChungJ-W, Wassle H. Electron microscopic analysis of the rod pathway of the rat retina./ Comp Neurol. 1993;332:421-432. Dacheux RF, Raviola E. The rod pathway in the rabbit: A depolarizing bipolar and amacrine cell./ Neurosci. 1986;6:331-345. Frederick JM, Rayborn ME, Hollyfield JG. Glycinergic neurons in the human retina. / Comp Neurol. 1984;227:159-172. Marc RE, Liu W-S. (3H) Glycine-accumulating neurons of the human retina. / Covip Neurol. 1985; 232:241-260. Strettoi E, Raviola E, Dacheux RF. Synaptic connections of the narrow-field, bistratified rod amacrine cell (All) in the rabbit retina. / Comp Neurol. 1992; 325:152-168. Lasansky A. Contacts between receptors and electrophysiologically identified neurones in the retina of the larval tiger salamander. / Physiol (Lond). 1978; 285:531-542. Djamgoz MBA, Kolb H. Ultrastructural and functional connectivity of intracellularly stained neurones in the vertebrate retina: Correlative analyses. Microsc Res Tech. 1993; 24:43-66. Nelson R, Kolb H, Freed M. Off-alpha and Off-beta ganglion cells in cat retina: I: Intracellular electrophys-

2403

The Proctor Lecture

55.

56. 57. 58.

59. 60. 61. 62. 63.

64. 65. 66. 67. 68.

69. 70.

71. 72.

73.

iology and HRP stains. J Comp Neurol. 1993; 329:6884. Kolb H. The organization of the outer plexiform layer in the retina of the cat: Electron microscopic observations. J Neurocytol. 1977; 6:131-153. Kolb H. The inner plexiform layer in the retina of the cat: Electron microscopic observations. J Neurocytol. 1979;8:295-329. Frederick JM, Rayborn ME, Laties AM, Lam DMK, Hollyfield JG. Dopaminergic neurons in the human retina./ Comp Neurol. 1982;210:65-79. Kolb H, Cuenca N, DeKorver L. Postembedding immunocytochemistry for GABA and glycine reveals the synaptic relationships of the dopaminergic amacrine cell of the cat retina. J Comp Neurol. 1991;310:267284. Kolb H, Cuenca N, Wang H, DeKorver L. The synaptic organization of the dopaminergic amacrine cell in the cat retina. J Neurocytol. 1990; 19:343-366. Pourcho RG. Dopaminergic amacrine cells in the cat retina. Brain Res. 1982; 252:101-109. Kolb H, Nelson R. Rod pathways in the retina of the cat. Vision Res. 1983;23:301-312. Kolb H, West RW. Synaptic connections of the interplexiform cell in the retina of the cat. J Neurocytol. 1977;6:155-170. Sterling P, Cohen E, Freed M, Smith RG. Microcircuitry of the ON-beta ganglion cell in daylight, twilight, and starlight. Neurosci Res. 1987;6(suppl):S269S285. Sterling P, Freed MA, Smith RG. Architecture of rod and cone circuits to the On-beta ganglion cell./Neurosci. 1988; 8:623-642. Hecht S, Schlaer S, Pirenne MH. Energy, quanta and vision./ Gen Physiol. 1942;25:819-840. Nelson R. All amacrine cells quicken the time course of rod signals in the cat retina. / Neurophysiol. 1982;47:928-947. Nelson R, Kolb H. Al 7: A broad-field amacrine cell of the rod system in the retina of the cat. / Neurophysiol. 1985;54:592-614. Nelson R, Kolb H, Famiglietti EV, Gouras P. Neural responses in the rod and cone systems of the cat retina: Intracellular records and procion stains. Invest Ophthalmol. 1976; 15:946-953. McGuire BA, Stevens JK, Sterling P. Microcircuitry of bipolar cells in cat retina. / Neurosci. 1984;4:29202938. Pourcho RG, Goebel DJ. Neuronal subpopulations in cat retina which accumulate the GABA agonist (3H)muscimol: A combined Golgi and autoradiographic study./ Comp Neurol. 1983;219:25-35. Cohen E, Sterling P. Microcircuitry related to the receptive-field center of the ON-beta ganglion cell. / Neurophysiol. 1991; 65:352-359. Kolb H. Cone pathways in the mammalian retina. In: Hilfer SR, Sheffield JB eds. Molecular and Cellular Basis of Visual Acuity. New York: Springer-Verlag; 1984:56-78. Miller RF, Dacheux RF. Synaptic organization and

74. 75. 76.

77.

78.

79. 80. 81. 82. 83. 84. 85. 86. 87.

88. 89. 90. 91. 92.

ionic basis of on and off channels in mudpuppy retina: III: A model of ganglion cell receptive field organization based on chloride free experiments./ Gen Physiol. 1976;67:679-690. Schiller PH. The ON and OFF channels of the visual system. Trends Neurosci. 1992; 15:86-92. Nelson R, Kolb H. Synaptic patterns and response properties of bipolar and ganglion cells in the cat retina. Vision Res. 1983; 23:1183-1195. Raviola E, Gilula NB. Intramembrane organization of specialized contacts in the outer plexiform layer of the retina: A freeze-fracture study in monkeys and rabbits. / Cell Biol. 1975; 65:192-222. Raviola E, Raviola G. Structure of the synaptic membranes in the inner plexiform layer of the retina: A freeze-fracture study in monkeys and rabbits. / Comp Neurol. 1982; 209:233-248. Nelson R, Famiglietti EV, Kolb H. Intracellular staining reveals different levels of stratification for oncenter and off-center ganglion cells in the cat retina./ Neurophysiol. 1978; 41:427-483. Hartline HK. The receptive fields of optic nerve fibers. Am J Physiol. 1940; 130:690-699. Kuffler SW. Discharge patterns and functional organization of mammalian retina. / Neurophysiol. 1953; 16:37-68. Mach E. The Principles of Physical Optics: An Historical and Philosophical Treatment. Anderson JS, Young AFA, trans. New York: Dutton. Werblin F. Synaptic connections, receptive fields, and patterns of activity in the tiger salamander retina. Invest Ophthalmol Vis Sci. 1991;32:459-483. Baylor DA, Fuortes MGF, O'Bryan PM. Receptive fields of the cones in the retina of the turtle. / Physiol. 1971;214:265-294. Enroth-Cugell C, Robson JG. The contrast sensitivity of retinal ganglion cells of the cat. / Physiol (Land). 1966; 187:517-552. Marks WB, Dobelle WH, MacNichol EFJ. Visual pigments of single primate cones. Science. 1964;43: 1181-1183. Maxwell JC. Experiments on colour, perceived by the eye, with remarks on colour blindness. Edinb Trans. 1855. Nathans J, Thomas D, Hogness DS. Molecular genetics of human color vision: The genes encoding the blue, green and red pigments. Science. 1986; 232:193202. Young T. On the theory of light and colours. Philos Trans R Soc Lond Biol. 1802; 92:20-71. Gouras P. Precortical physiology of colour vision. In: Gouras P ed. The Perception of Colour: Vision and Visual Dysfunction. London: Macmillan; 1991; 6:163-1 78. Gouras P. The History of Colour Vision. In: Gouras P, ed. The Perception of Colour: Vision and Visual Dysfunction. London: Macmillan; 1991; 6:1-9. Boycott BB, DowlingJE. Organization of the primate retina: Light microscopy. Philos Trans R Soc Lond Biol. 1969;255:109-184. Boycott BB, Wiissle H. Morphological classification of

2404

Investigative Ophthalmology & Visual Science, April 1994, Vol. 35, No. 5

bipolar cells of the primate retina. Eur J Neurosci. 1991;3:1069-1088. 93. Mariani AP. The neuronal organization of the outer plexiform layer in the primate retina. Int Rev Cytol. 1984;86:285-320. 94. Rodieck RW. The primate retina. In: Comparative Primate Biology. Vol. 4, Neurosciences. New York: Alan R.

Liss; 1988:203-278. 95. Kolb H, DeKorver L. Midget ganglion cells of the parafovea of the human retina: A study by electron microscopy and serial section reconstructions./ Comp Neurol. 1991; 303:617-636.

96. Gouras P. Identification of cone mechanisms in monkey ganglion cells./Physiol Lond. 1968; 199:533-547. 97. Kolb H. The neural organization of the human retina. In: Heckenlively J, Arden G. Principles and Practice of Clinical Electrophysiology of Vision. St. Louis: Mosby

Year Book; 1991:25-52. 98. Kolb H. Anatomical pathways for color vision in the human retina. Vis Neurosci. 1991;7:61-74. 99. Kolb H, Lipetz LE. The anatomical basis for colour vision in the vertebrate retina. In: Gouras P, ed. The Perception of Colour: Vision and Visual Dysfunction. Lon-

don: Macmillan; 1991:128-145.